Archeologia e Calcolatori 5 - 1994

Principal component analysis (PCA) is a widely used multivariate method in archaeology, and is particularly prevalent in archaeometric applications. The paper reviews the use of the methodology in archaeometry, including the choice of data transformation and standardisation. The related methods of factor analysis and correspondence analysis are also briefly considered. Two detailed examples illustrate some of the methods discussed, including uncommon approaches such as the use of ranked data.

Results of the archaeometric study of Spanish amphoras in Veneto are presented. For this purpose 26 amphoras from the Tomba di Giulietta in Verona were chosen (types close to Dr 7 and Dr 8). The scope of the analytical study was to determine similarities between clays and thus between the amphoras themselves. To conduct petrographic analysis, groups were set-up of Spanish amphoras of both types without determining relationships between various forms and clay bodies. North Italic amphoras were also considered for comparison. The most useful methods of statistical analysis of data of diffractometric analysis are “ipercubo” and Discriminant Analysis. Fuzzy and mixture methods were also used. Applying these methods, samples were compared in order to form groups of similar samples. Instrumental neutron activation analysis was carried out to determine trace elements and the results were processed with Discriminant Analysis. These results enabled us to make clear distinctions between various groups. Studies of these amphora forms are now under way.

In recent years, archaeologists have become increasingly aware of the added interpretability that can be gained if they adopt a model-based approach to their statistical analyses. In this paper we encourage the use of just such an approach for the investigation of dimensional data from the Sardinian corbelled structures known as Nuraghi. We highlight the fact that the experts who surveyed these domes have gained extra information (other than the basic structural measurements) which is not usually included in statistical investigations in any formal manner. Here we advocate the adoption of the Bayesian statistical framework within which the model, the data and the expert knowledge can be combined in a consistent manner and thus all permitted to influence our final inferences. By using examples we are able to demonstrate that, for at least some of the domes surveyed, the inclusion of the expert knowledge has a marked effect on the result obtained.

The aim of the study is to observe space and time variations in the occupation of Sicilian funerary settlements from the 9th to the 1st century BC The data have been extracted from a complete examination of archaeological reports. They define geographical and archaeological characteristics of 240 settlements. From archaeological to statistical data there is an alteration of variable status which represents a new field for archaeologists. The matrix summarises the settlements’ occupation in 50 year periods by presence-absence; this gave rise to a Correspondence Analysis. The time of occupation has been preferred to a relative chronology based on cultural facies. One additional variable distinguishes the cemeteries where the graves with several burials dominate over those having individual burials. The results of the analysis prove that the space and time variations in the two types of funerary settlements described oddly follow the outline of historical events in relation to autochthonous and colonial spheres. In particular it shows the alterations of autochthonous funerary customs owing to more archaic cemeteries which were to be reused later on. In the historical sphere, the analysis leads us to question the acculturation phenomena. What was happening in Sicily towards the beginnings of 5th BC? Did autochthonous funerary customs disappear because of the adoption of Greek culture, or is it getting more difficult to establish a difference from the ancient colonial culture?

The article presents problems connected to the management of iconographic data using computer science tools. In particular, the authors discuss especially the computerised management of Athenian figure-decorated pottery which represents perhaps the most relevant set of iconographic patterns in classical antiquity. Up to now this pottery has been investigated by means of quantitative analyses to study in particular the mechanics of ancient trade in the Mediterranean basin because Athenian vases, from 6th to 4th century BC were widely exported all around this area. The computerisation of iconographic information, however, presents many difficulties: in fact archaeologists tend to express iconic information by means of linguistic sentences which are often inadequate; moreover there are still many problems about the correct interpretation of representations on attic pottery. On this subject, the authors illustrate a project on Athenian decorated pottery found at Spina, the Etruscan “emporion” of the Upper Adriatic, where the most impressive collection of Athenian decorated vases of the 5th century BC was discovered. A database has been created which collects data on 2132 clay decorated vases made in Athens: iconographic information is represented both with a description in natural language and by means of terms of a thesaurus conceived for this purpose. The database has also been explored through statistical analyses, such as Correspondence Analysis and Multidimensional Scaling, to find significative interrelationships between shapes and subjects, to identify associations between the different subjects represented on the same vase and to stress changes, as time passed, with regard to shapes and subjects.

Commencing with a review of the application of information processing to epigraphy, the use of image analysis and pattern recognition for the identification of epigraphic workshops is proposed. The project consists of the following steps: a) the collection of coeval documents from one and the same region (e.g., an ancient community); b) the extrapolation of single characters to be superimposed on the same texts in order to proceed to a statistical analysis of variations; c) the development of an “average” pattern; d) the superimposition (manual or automatic) of the pattern within the single texts, so as to ascribe them to the different workshops. An application of the method by the Centre for Textual Studies, Oxford, is described. The application allows rapid access to the PHI CD-ROM, developed and distributed by the Packard Humanities Institute, Los Altos, California.

CThe “Giunone” project concerns the implementation of a numismatical database, designed to become a valid tool for researchers in the field of numismatics and coin collection history. The database was built in conformity with the entity-relationship model and is used in this phase under a Unix System. The task of data entry can be carried out by different people using a programme that works on IBM computers.

This paper begins with four propositions: there are four basic types of information: data (or more precisely numbers), text (i.e. single words, sentences and longer compositions), images (still and moving), and sound; that computer technology has now advanced sufficiently in power to capture, store, manipulate and retrieve all the types of data listed in the first proposition; that archaeological information is composed of all five types of information outlined in the first proposition; and that since archaeology also has vast quantities of this information it needs computer technology in order to handle it successfully. Given these propositions, this paper describes the purpose of the Ashmolean Museum of Art and Archaeology and defines three separate roles that result. The historical development of IT applications since the late 1970s is traced, highlighting the mistakes made and the lessons learnt. This culminates in a description of the Ashmolean’s Collections Information System, in particular outlining the structure of the Collections Information Database that underpins the system. The plans for implementing this system are also described. The paper ends by looking at the problems of fully realising this system, and concludes that the plans are really a means of deliberately staying several steps behind the full potential of the technology while still moving forward, so that the Museum can wait for the eventual arrival of software that can provide a seamless transition between various tasks, with a minimum of effort.

The potential of the relational database management system (RDBMS) as a means of integrating disparate archaeological data sets is explored, using data recovered from the stratigraphically complex environment of excavations within the city of London. Means of checking and modelling the single context stratigraphies from such sites within the RDBMS are considered, followed by descriptions of techniques developed to allow integration of that model with dating information in order to aid identification of residual and intrusive material. Relational databases are seen to be a realistically accessible technology for archaeological bodies, and are of especial relevance to those working in the increasingly competitive commercial environment of developer funded rescue excavation, as is now common in London and England in general. They offer a means of aiding interpretation through both automating some of the repetitive tasks of site analysis and more importantly by allowing the wider patterns extending across previously discrete data sets to be sought.

Computerised databases form an integral part of much archaeological work around the world, but few of these systems are designed with expansion or compatibility as a prerequisite. This paper investigates the problems of integrating data from a variety of sources, both urban and rural, and then discusses in detail the modular database solution adopted by two archaeological projects at the University of York. The discussed solution enables the integration of data collected from different sources, and at very different scales from the single record per site to the recording of every feature or artefact in more detailed databases. To conclude, the paper touches upon future developments in the field, and their potential impact upon the way in which we store and view archaeological data.

Archaeology is starting to face the challenges of electronic publication posed by advances in information technology; we are now able to offer almost all types of archaeological data in a single software environment, and have the opportunity to integrate hitherto discrete results. This paper describes the use of the University of Southampton’s ‘Microcosm’ open hypermedia system in an archaeological context. Microcosm is a set of programs running under Microsoft Windows which can support a variety of archaeological data, including databases, digitised site plans and survey data, synthetic text, and GIS images. The example uses the data generated by the Urban Origins in East Africa project, a Swedish-funded initiative to bring together archaeologists from ten East African countries and look at a regional problem. We suggest how the integration provided by multimedia can not only be a vital part of research work, but also be used in publicity, and provide the structure of a genuine regional resource.

The author illustrates the structuring that was carried out for the development of the ArchéoDATA project with the aim of forming a basis for a more general Archaeological Information System. The conception of this project was undertaken from an archaeological perspective; the subsequent use of computers, and in particular the possibilities offered by GIS, enabled the enhancement of the system. The problems of basic information are also discussed, with an attempt to overcome the traditional concepts of site, finds and features by replacing them with the more general concepts of spatial and temporal entities and interpretative grouping.

The authors illustrate the application of a GIS technology within the field of historical archaeology. They discuss a GIS pilot study of the Adriatic island of Hvar in Croatia, using the Geographical Resource Analysis Support System (GRASS), a Unix based software. Basic data was obtained through a survey of all the archaeological monuments on the island. In contrast to Greek literary sources, the GIS application gives a different perspective of the historical Greek colonisation of the island of Hvar (385/4 BC). In fact, after founding Pharos, there is no evidence of permanent Greek settlements extending beyond the plane, so the life of native communities would have continued unchanged.

This paper is a synthesis of an experimental study on ancient topography and town-planning in a historical quarter of Rome, Trastevere. The focus is on methodological aspects, and in particular on computerized applications. Thy have been used, at first, to speed up the input and the management of data, but the results deeply influenced the methods and the logical course of the research itself. The study can therefore be considered also as a proposal for a modular application to other urban contexts. Traditional sources of information were taken into account, as in any study on ancient topography, including the archaeological map (concerning both visible and bibliographically documented elements) drawn on the basis of present-day cartography, literary and archival sources and historical cartography (from Forma Urbis Severiana to Catasto Pio-Gregoriano). In addition to the study of some important methodological consequences of this approach, the results are outlined of having applied computerised procedures of cartographical transformation, not previously adopted in any archaeological framework.

The Geohm System belongs to the geo-electrical family of prospecting devices, and is a system for substratum exploration by electrically scanning along lines of preplanted sensors. The system generates a horizontal section of the values of resistivity in archaeological sites. The Geohm is an uncomplicated system comprising of a portable computer, an analogical-digital conversion device, a multi relay switcher (software controlled), a solid state current converter (also software controlled), and a lot of moving sensors inserted in the ground. The measured groups of returned signals are processed by complex algorithms before the data is validated: this technique makes it possible to obtain more reliable measurements and allows the user to reprogram the device. The speed of the system allows the user to survey a large area using several electrical devices.

S. Rosa terramara, which measures 7 hectares, is a Middle and Recent Bronze age site, excavated since 1984. Cooperation between archaeologists and geologists has made it possible to experiment with the comparative analysis of computer image processing of the aerial photos, integrated with the detailed study of a DTM (Digital Terrain Model), in particular in relation to the geopedological and geoarchaeological interpretation of the site. The integration of two kinds of data, aerial photographs and a DTM, has been complemented by the texture mapping of the aerial photo overlaid on the corresponding 3D model. Significant ranges in the pixel distribution in the digital aerial image identified four main areas; any area presents particular digital features. For example, the digital features of the small village are very different from the characteristics of the large and more recent village. The digital classification of these areas identified the following: area 1, alluvial sediment (posthumous) which hides part of the earthwork; area 2, the nucleus of the small village; area 3, the diking of the big village; area 4, a Roman villa (another posthumous event). The final interpretation, after processing, distinctly shows the formation of natural (vertisoils) and artificial deposits (earthworks). In particular in the north-western area of the aerial photograph the interruption of the diking means that in that point the structures are buried because of a flood. Moreover, it results that the two villages are separated by a large ditch. On the basis of these results it has been possible to reconstruct the evolution of the site and of the environment in different periods: 1) the site before the Bronze age; 2) the terramara in the Middle Bronze age; 3) the terramara in the Recent Bronze age; 4) the site in Roman age; 5) the site in Medieval age. The computer processing, integrated with the aerial photo-interpretation, shows an important series of data not obtainable through traditional techniques. The construction of the DTM, the texture mapping and the digital image processing have clearly enhanced the shape of the structures of the terramara as well as the stratigraphical excavations. This also resolves the problems which emerged in the aerial photo interpretation. In particular, the interruption of the earthwork of the big village in the northern area seems connected with the burial of part of the anthropic structures, after the abandonment of the terramara. In conclusion, the computer image processing, together with the DTM and the texture mapping of the site, is an exportable tool, useful for an evaluation of the state of preservation of the Bronze age deposits in the Po valley. These results were obtained without excavation, using only simple stratigraphical surveys on the ground.

Ceramigraph is an electronic and mechanical system designed to draw pottery through automatic processing and therefore excluding the problems of manual drawing. The system, particularly suitable for large quantities of pieces, is composed of a high precision 3D pantograph, connected with a Personal Computer and a digital tablet. The main features of the drawings are the following: digital; vectorial; processed by software; classifiable (on the basis, for example, of the Freeman code); printable on normal or glossy paper at any scale; recordable on databases, CD rom, syquest, floppy disks, and so on (the size of vector files is very small; for example on a 1.44 floppy disk, 100 drawings can be recorded). The classification and the representation of the graphical information processed in this way are fundamental for the typological and typometric interpretations.

The authors show here, through an example, how the reasoning procedures in an archaeological construct can be analysed, represented, critically evaluated, and possibly transcribed so as to be used again as part of an expert system. The monograph which has been analysed is an extract from a study on the prehistorical origins of the myth of Classical Greek Demeter. The reasoning steps of the author are first extracted and rewritten in the form of chains of inferences, according to J.-C. Gardin’s logicist approach. These components are then transcribed in terms of a fact base and a rule base according to the SNARK system. The SNARK knowledge base, as well as the results of the computing, are presented as tools for a better understanding of reasoning in the humanities.